Out of curiosity, why do you make this statement about cartridges in the 22 to 28 grain range?
The key words are 'small cartridges'. It's a reflection of the size of powder charge that sees different primers produce different outcomes. It presumably affects other calibres too with small enough charges and cases, eg 30BR, but I've no experience here.
The other thing to note is that while a straight primer substitution may (usually does) affect short-distance precision that still doesn't in itself prove that one model is better or worse than another as subsequently refining the powder charge weight often compensates. The important issue is not to regard primers as interchangeable so far as precision goes, and ideally once a good load is worked up stick with that primer, better still the original primer production lot, if consistently producing small groups is the shooter's primary concern. (I have also on occasions found that a single primer model is head and shoulders above the rest of the pack with a good-performing bullet/powder combination in these small cartridges, and the obvious conclusion is to get hold of enough of that primer model to last the rifle's barrel out. I imagine that this effect is what Bill Alexander is referring to in his Grendel primer choice advice.)
ES/SD values are often affected too by primer substitution, but at short distances don't in themselves affect short-distance group size, or if there is a direct relationship it's one I've never encountered. In fact, there almost seems to be a perverse link in that the number of times the powder charge that gives the smallest group also gives the largest ES (or it produces small groups and ES values but at 150fps too low MV). In an ideal world, one wants the 'Goldilocks' combination that gives all things for longer distance shooting - high enough MVs, low SDs, and superb precision.
The 223, and I don't suppose the AI variant is any different, is a sod to get small spreads out of. There is at least one live thread running on the forum as to why this should be so. Look up
@Ned Ludd 's posts on this as well as other issues of making 223 shoot well at long distances with heavy bullets. In my case, any ES below 20 fps is regarded as acceptable with this cartridge and that involves powder charge variations smaller than plus or minus 0.1gn. (As 0.1gn changes 223 MVs by typically 10-12 fps with 80gn and heavier bullets, sometimes more depending on powder, a 0.2gn charge weight range induces
at least 20-25 fps ES before any other effects kick in as they do with any cartridge. Whether you need to have charge consistency down to the single powder kernel, ie ~0.02gn, depends on the combination of needed precision and distance involved - 1,000 yard F/TR shooting on targets with a MOA 10-ring and half-MOA X-ring is a
very demanding application for the 223 even with 90/95gn bullets. 100 or 200 yards, even with out and out precision needed, much less so at least in this respect.